It is an exciting time for those who share the vision of a patient-centric, womb-to-tomb electronic health record (EHR). At the core of that vision is true portability of care with private, secure, on-demand access by health care providers to a patient’s complete medical history via a national health information network (NHIN) not unlike those forming in Canada, the UK, and other developed nations.

Across the United States, interest is surging in establishing regional health information networks (RHINs) intended to exchange digital patient data across physical and organizational borders. Financial, legal, and organizational support is growing at federal and state levels, and several promising regional pilots are under way.

To date, however, diagnostic imaging providers have played a minor role in this movement despite the benefits such networks could provide to radiologists, cardiologists, and surgeons in the form of enhanced access to a patient’s historical images. Moreover, imaging’s technical and financial characteristics make it the ideal starting point for cross-enterprise exchange of clinical information.

RHIOS AND RADIOLOGY

Thus far, most efforts at cross-enterprise exchange (increasingly in the form of regional health information organizations or RHIOs) have focused on sharing laboratory, pharmacy, or narrative clinical data. A few have even periled big bang attempts to share all forms of clinical data across their region at launch time.

All of these initiatives have in common a continuing struggle with three critical obstacles to their success: 1) financing the deployment of information technology (IT) among sites still mired in paper, 2) ensuring interoperability among participants, and 3) establishing sustainable business models for the launch, operation, and support of such networks.

While intuition suggests that access to out-of-reach clinical data will improve diagnostic accuracy, timeliness, and clinical outcomes, studies intended to demonstrate and quantify such dividends remain in their infancy. In their absence, the business case for financing regional networks remains rooted largely in the elimination of administrative overhead, which, in many cases, appears insufficient for initial construction and ongoing operation.

RHIOs or other organizations seeking to rapidly launch regional networks at minimal cost and with maximal impact on the quality and cost of care would do well to consider leveraging imaging’s years of IT leadership, not to mention the millions of dollars already invested in information systems.

STRONG CLINICAL SIGNIFICANCE

To begin with, regional exchange of imaging data supports a central tenet of clinical practice: side-by-side comparison of new images with a patient’s historical ones. Such comparisons are known to have a dramatic impact on diagnostic accuracy and increased physician confidence. 1,2

Elliot D. Menschik, MD, PhD

Despite being the standard of care, 3 radiologists often do without the aid of priors, whether screening for breast cancer or evaluating a solitary pulmonary nodule. Recent data suggest that, in approximately 20% of cases, radiologists ask for but are unable to access relevant priors due to their location beyond the borders of their own institution. 4

Now imagine a picture archiving and communications system (PACS) that prefetches and renders not only priors from the local archive, but from the heterogeneity of PACS at other sites in the region or across the nation. This is technically feasible and realizable today, and regional exchange of imaging across institutional borders should be expected to have a direct and measurable impact on improved outcomes while simultaneously mitigating against malpractice risk. 5

IMAGES ARE DIGITAL AND STANDARD

From a technology perspective, an initial focus on imaging makes it possible for RHIOs to launch fully digital and interoperable networks today, not 10 years from now as is assumed to be the timeline for widespread adoption of electronic medical records (EMRs).

As we wrestle with deploying EMRs across a nation whose health care delivery remains firmly tied to paper, imaging has spent the last two decades moving to digital acquisition, storage, distribution, and interpretation of clinical data. As a result, where perhaps 10% of all clinical data is in some digital form today, approximately 45% of the estimated 500 million imaging procedures performed annually are archived digitally within the walls of hospitals, practices, and freestanding imaging centers. All that is lacking is the connectivity a regional network can provide.

While new standards are laying the foundation for data portability across all specialties (eg, the Continuity of Care Record and HL-7’s Clinical Document Architecture), their widespread implementation in production-level clinical environments remains years away. In contrast, the DICOM standard is both battle-tested and near universal implementation. While not perfect, DICOM provides for a high level of vendor-neutral interoperability in the real world. Moreover, DICOM is now being harnessed to the needs of regional data exchange by the Cross-Enterprise Document Sharing (XDS) integration profile to be released this year by the Integrating the Healthcare Enterprise (IHE) initiative.

IMAGING As FINANCIAL BOOTSTRAP

Perhaps most important, launching regional networks with an initial focus on imaging provides for immediate and sustainable financing through the reallocation of funds already committed to film by network participants.

Specifically, distributing images across institutional borders is not only common, but is responsible for significant expenditures even at PACS-enabled institutions. For example, the Hospital of the University of Pennsylvania has reported approximately $2.5M annually in film expenditures even 8 years after moving to PACS. 6 This expense has been driven in large part by the need to communicate results with referring physicians in the community who are not formally affiliated with the health system and will likely never have a virtual private network (VPN) account or other direct access to the system’s own network.

Unlike other forms of clinical data that are routinely communicated across borders by phone or facsimile, such transmission of images today depends largely on film. Regional exchange of imaging data has the potential to eliminate this film burden, translating to multiple millions of dollars saved.

In short, regional health information networks can provide the “last mile” to filmless operation, and the dollars already budgeted for film expense by potential network participants can be operationalized to finance the launch and ongoing operation of such networks. Once the technical and support infrastructure has been built on film-derived financial savings, network expansion to other forms of clinical data becomes incremental and easily absorbable costs.

CENTRALIZED Versus DISTRIBUTED

For those looking to share images between institutions and/or collaborate on care across enterprise borders, network architecture is a critical consideration. Given a choice between centralized and distributed storage models, several key factors point to distributed networks as best suited for imaging data if not all forms of clinical data as well.

Given that an average metropolitan region today generates on the order of 40TB of images annually (and ever increasing as next-generation imaging modalities provide ever-growing data sets), networks built around a shared central archive have both immense storage and bandwidth requirements (ie, 100% of data volume must be transmitted to the shared archive, plus another 20% or so might be expected to be retrieved at a later date).

In contrast, a distributed model leverages existing investments in clinical IT, leaving data at the source site, but enabling on-demand access to the data by other network participants as appropriate. Not only does this approach remove the need for costly central storage, but it cuts the required bandwidth by approximately 80% relative to the centralized model.

Centralized vs. Distributed Network Architectures

As important as these technical considerations are in light of imaging’s large data sets, the network architecture also has important ramifications for the political and procedural dimensions of regional data exchange. A central model requires agreement across all participants on who “owns” the data and on the establishment and enforcement of common data privacy and security policies. While these are obstacles that can be addressed by a RHIO in its internal deliberations, it is worth noting that these issues were at the heart of the failure of the community health information networks (CHINs) of the 1980s and 1990s. CHINs collapsed amid disagreements linked to the centralized model of community sharing.

In contrast, the distributed model renders moot a discussion over ownership of clinical data by retaining the status quo, with each site continuing in its role as steward of its patients’ data. Moreover, the distributed model provides much greater flexibility in the implementation and enforcement of privacy and security policies, enabling each site to maintain its independence and reproduce the policies and procedures already in place for managing the disclosure of analog data to outside entities.

PHILADELPHIA EXPERIMENT

These ideas are being put into practice by an ambitious health information exchange linking health care providers in the Philadelphia region. Focused initially on sharing diagnostic-quality digital images and reports, a promising collaboration and network have sprung up among some of the region’s key provider organizations including the University of Pennsylvania Health System, Thomas Jefferson University Hospital, and Children’s Hospital of Philadelphia.

Initially funded by the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the network currently links heterogeneous PACS and RIS across multiple sites and manages some 140 million images covering 8 years of patient history.

While expansion to additional sites in the region is planned via the bootstrapping method described above (ie, real location of film expense), the National Cancer Institute (NCI) is generously supporting research into the network’s impact on cancer care, enabling diagnostic workstations to directly access imaging data from other sites (as opposed to web access alone), and upgrading the network to enable compliance with IHE’s XDS profile and other relevant open standards.

In many respects, if such a network can exist and thrive in a multi-vendor, high-volume, competitive environment such as Philadelphia, it should prove a potent model for other regions seeking to take the next step in the evolution of medical care.

Centralized vs. Distributed Network Architectures

WHERE TO START

Providers seeking to realize the clinical and financial benefits for their practices and patients can take concrete steps today that can lead to a production-level network within a matter of months.

The first step is to find and convene like-minded leaders from other regional hospitals and practices who will each champion the regional exchange within their own organization. A key role here is the ability to loop in the right local team members as appropriate from a chief information officer (eg, to guide/customize a technical approach) to general counsel (eg, to clarify existing privacy and security policies). While the creation of a formal RHIO as a stand-alone legal entity is not required, it should at least be considered as a means of leveling the playing field among multiple organizations, providing clear rules of governance, and eventually taking advantage of regulatory benefits targeted specifically to RHIOs (eg, changes in the Stark and/or federal anti-kickback statutes).

Second, the cross-enterprise team should develop an estimated budget for the endeavor. In the case of imaging, this is quite straightforward as each team member can examine their own film-based expenditures and estimate how much is spent a year on distributing film to outside referring physicians. While an exchange will not immediately eliminate all of this expense, a reasonable estimate can be made of the fraction of film funds that can be freed and put to work launching and maintaining the exchange.

Third, the team should decide whether to build or buy. While some centers may have the in-house expertise to develop, deploy, and support a regional network, this is likely to take longer to realize than desired while straining technical development and support capabilities that are likely to already be at their limits. In essence, this approach misses out on the expertise and economies of scale that commercial service providers are now bringing to the table.

For those who opt to buy, the market for regional health information solutions has matured rapidly over the last couple of years, and today there are essentially two choices: 1) a long-term engagement with a systems integrator (SI)/consulting firm to plan, develop, implement, test, debug, deploy, and support a network, or 2) evaluation and selection of a scaleable, customizable, and flexible drop-in solution from one of a handful of dedicated “health information service providers.” HISPs are a relatively new category of vendor dedicated to partnering on regional exchanges and with expertise born of connectivity and integration projects within and, more recently, between integrated delivery networks. In contrast to the consulting/SI approach, HISPs will have a ready-to-go solution that can be customized to the needs of the regional participants rather than built from scratch.

While hiring a traditional SI firm might be a better fit for institutions more comfortable with established brands, HISPs are likely to have already worked through the technical challenges and be able to provide a modular solution that can be deployed more rapidly and at lower cost, and has already been proven in one or more markets. Such solutions are also more likely to accommodate new sites that want to join the network further down the road.

These choices aside, imaging sits in a unique position in the clinical spectrum by virtue of its digital and standard nature, the diagnostic importance of historical images, and the continued significant expense of film in a post-PACS environment. Imaging providers can and should take advantage of these characteristics today to establish regional exchanges that extend the abilities of their current PACS/RIS investments, maximizing clinical benefits for their patients and financial gains for their practices.

Elliot D. Menschik, MD, PhD, is the founder and chief executive officer of Hx Technologies (www.hxti.com), a Philadelphia-based health information service provider.

References:

  1. White K, Berbaum K, Smith WL. The role of previous radiographs and reports in the interpretation of current radiographs. Invest Radiol. 1994;29:263-5.
  2. Aideyan UO, Berbaum K, Smith WL. Influence of prior radiologic information on the interpretation of radiographic examinations. Acad Radiol. 1995;2:205-8.
  3. Standards. Reston, Va: American College of Radiology; 1995.
  4. Lakhani P, Menschik ED, Goldszal AF, Murray JP, Weiner M, Langlotz CP. Development and validation of queries using structured query language (SQL) to determine utilization of comparison imaging in radiology reports stored on PACS. J Digit Imaging. In press.
  5. Berlin L. Must new radiographs be compared with all previous radiographs, or only with the most recently obtained radiographs? AJR Am J Roentgenol. 2000;174:611-5.
  6. Goldszal AF, Bleshman MH, Bryan RN. Financing a large-scale picture archival and communication system. Acad Radiol. 2004;11:96-102.
  7. Lorenzi NM. Strategies for creating successful local health information infrastructure initiatives. Vanderbilt University. Report #03EASPE00772. December 16, 2003. Available at: http://aspe.hhs.gov/sp/nhii/LHII-Lorenzi-12.16.03.pdf.